Many systems and processes exhibit non-linear behavior that must be taken into consideration. Furthermore, real systems have uncertainties and time-variant behavior, the effects of which must be compensated by the controller.
Various methods from automatic control and artificial intelligence address these requirements. Model-predictive control can deal with non-linearities in the system and allows the formulation of optimization-based control objectives. Robust and adaptive control is based on a model structure and considers model uncertainties and unknown, time-varying parameters. Data-driven methods are based on the input-output behaviour of systems and require only little model knowledge.
At the Chair of Automatic Control two research projects deal with these topics:
- Control of physical properties during free-form bending processes